As radiology departments evolve toward managing images in a digital format, image compression techniques are being advocated as a means to alleviate problems caused by the large volume of digital data required to represent each image. This issue is of particular importance in mammography. The value of mammography in the early detection of breast cancer is widely recognized. At the same time, because of spatial resolution and contrast sensitivity requirements and other image quality considerations in mammography, mammograms present unique challenges and opportunities with respect to the application of image compression technology. Nevertheless, in view of the recent adoption of a general purpose image compression standard by the Joint Photographic Experts Group (JPEG) of the International Organization for Standardization (150), we believe that it is crucial at this time to assess the applicability of the standard algorithm to the compression of mammographic images. In this project, we are proposing to optimize the standard algorithm for the compression of digitized mammograms and then to measure the dependence of diagnostic performance on the degree of compression for this optimized algorithm. The study design will enable us not only to measure differences in performance by paired comparisons, but will also define trends, if they exist, and through regression procedures will permit us to evaluate the "compression/performance" curve for mammography.